Mapping Preferences: A Harry Potter Topology
Clusters in a graph visualization are collections of closely related nodes that share many of the same links. Being able to easily see clusters is helpful for showing relationships. But clusters don't always stand out easily in a node-link graph, even when nodes are color-coded. Besides, many people are not familiar with the concept of a graph.
To better visualize graphs, a new algorithm, GMap, is being developed in Graphviz. It uses the geographic map metaphor, highlighting each cluster in a differently colored region (or country, to stay with the map metaphor). The borders of the cluster countries are drawn half way between two nodes. Nodes within a country are much more closely related than nodes in bordering countries. Compared with node-link graphs, maps are much more familiar and intuitive to people.
Showing clusters on a map is also an interesting way to present lists of preferences such as those that maintained by Netflix or Amazon to suggest other selections based on common preferences. But sometimes a simple list of recommendations can seem odd or counter-intuitive when the overlap between two selections isn't clear.
Visualizing and mapping recommendations can remove some of this mystery, showing how a recommendation is grouped with other choices. A map, much more than a simple list, can also lead a user's eye to other options and encourage the user to explore the landscape.